You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'll see if I can find to make an MRE later, as right now I have spent too much time on this.
Are there any known quirks when working with the ICP function in CloudComPy compared to using the CC native GUI? Or should a trivial opening of the two files and calling into the function with the defaults match almost exactly opening the two files in the GUI and using the ICP tool, also with the UI defaults?
I am simply loading in a cloud and a reference mesh, aligning as per the ICP tutorial in the docs, cloning the aligned result, applying the matrix, and then saving the lot to file, which I then inspect in ccViewer - nothing complicated thus far. Today I've modified the cloudcompare GUI build to trace all the arguments passed to the ICP implementation, and what I see logged is identical to what I am calling into from Python (most are at their defaults anyway). Whilst the GUI aligns the two pretty well, the matrix that comes out of the Python call is off by about 30 degrees and barely an improvement over the starting transformation between the two.
Are there any obvious causes of something like this? Or is it possible/expected that the GUI and Python results may differ without some pre-processing or careful pinning of circumstances?
Thanks.
The text was updated successfully, but these errors were encountered:
Hi,
I'll see if I can find to make an MRE later, as right now I have spent too much time on this.
Are there any known quirks when working with the ICP function in CloudComPy compared to using the CC native GUI? Or should a trivial opening of the two files and calling into the function with the defaults match almost exactly opening the two files in the GUI and using the ICP tool, also with the UI defaults?
I am simply loading in a cloud and a reference mesh, aligning as per the ICP tutorial in the docs, cloning the aligned result, applying the matrix, and then saving the lot to file, which I then inspect in ccViewer - nothing complicated thus far. Today I've modified the cloudcompare GUI build to trace all the arguments passed to the ICP implementation, and what I see logged is identical to what I am calling into from Python (most are at their defaults anyway). Whilst the GUI aligns the two pretty well, the matrix that comes out of the Python call is off by about 30 degrees and barely an improvement over the starting transformation between the two.
Are there any obvious causes of something like this? Or is it possible/expected that the GUI and Python results may differ without some pre-processing or careful pinning of circumstances?
Thanks.
The text was updated successfully, but these errors were encountered: